Interested in joining the Silph Research group? Learn More »

Silph Study: #024

Published:
05.21.2019

Breakthrough: How Guaranteed Lucky Trades Actually Work

The Silph Research Group’s long-awaited article on the mechanics of Lucky Trades caused a stir when it reported trading Pokémon caught in July and August 2016 (by researchers who had each already received 10 Luckies) resulted in Lucky Pokémon in 75% of trades. Many travelers reported that the Silph Research Group’s findings did not appear to match their observations. While it would be tempting to suspect reporting bias, the Research Group’s datapoints were collected in a highly controlled setting. We suspected more was going on under the surface than met the eye.

To get to the bottom of these reports of varying rates, the Research Group put the mechanics of “guaranteed Lucky Trades” under the microscope. As it turns out, guaranteed Lucky Trades don’t operate quite the way they were announced! Here’s what we learned…

How Everyone Thought Things Worked

When Niantic announced changes to the Lucky Trade system last September, including the addition of guaranteed Lucky Trades for some trades involving Pokémon caught in July or August 2016, they stated the following:

“If you and your friend have received 10 or more Lucky Pokémon, the guarantee won’t be in effect, so keep that in mind! Now is the perfect time to bring old friends back to Pokémon GO and work together to complete your Kanto Pokédex!” – source (pokemongolive.com)

Travelers quickly found that this 10-Lucky limit applied to the sender of the guaranteed Lucky Trade (i.e., the trainer sending the July-August 2016 Pokémon) and that the trading history of the receiver of the July-August 2016 Pokémon did not appear to affect the Lucky guarantee.

Consequently, researchers were required to record (for each trade) both their own and their trading partner’s previous total lifetime Lucky Trades. This was done in order to control for guaranteed Lucky Pokémon in the Research Group’s investigation into Lucky rates. This record-keeping was intended to keep any guaranteed Lucky Trades from contaminating the analysis of non-guaranteed trades.

Everything appeared to be controlled for. However…

How The Guarantee Actually Works

A piece of the puzzle was missing. Now, thanks to a series of carefully controlled experiments conducted by Silph Research Group Scientists, as well as Lead and Senior Researchers, we can report that at this time1 the actual mechanism for guaranteed Lucky Trades involving a July-August 2016 Pokémon appears to be:

All July-Aug 2016 Pokémon are guaranteed Lucky if the sender has been in fewer than 10 trades (as sender or recipient) where at least one party was guaranteed.

At first, this may seem like a subtle distinction, but it means that not all Lucky trades are created equal. Lucky trades that were not guaranteed before the total is met do not appear to count towards the 10 guaranteed allotment.

Ignorance of this previously-unknown distinction meant many of the data points previously believed non-guaranteed were in fact guaranteed!

Testing Methodology

In order to test this hypothesis, the Silph Research Group tracked down a group of seven retired, returning or casual trading trainers2 (we’ll use Ash, Misty, Brock, May, Dawn, Gary, and Kiawe as pseudonyms for their trainer names), whose accounts had not previously been involved in any Lucky Trades involving July-August 2016 Pokémon. All trainers except for Misty and Gary had not previously made any trades at all.

In our first experiment, Dawn and May traded new Pokémon with several auxiliary trainers (i.e., not the trainers named above) until they each obtained a total of 10 Lucky Pokémon. These Lucky Trades were all non-guaranteed since none of the trades involved any Pokémon caught in July-August 2016. Under the old understanding of the guaranteed Lucky mechanism, neither Dawn nor May should have been able to send a guaranteed Lucky Trade after this point.

May then traded five July-August 2016 Pokémon to Misty in exchange for new Pokémon, and all five trades became Lucky. Next, Dawn traded ten July-August 2016 Pokémon to May, and all of them became Lucky. An eleventh July-August 2016 Pokémon sent from Dawn to May was non-Lucky.

That the first 15 of these trades became Lucky strongly suggests (though of course does not prove) that these trades were indeed guaranteed to be Lucky. Even at the disputed 75% Lucky rate, the probability of obtaining 15 consecutive Lucky Trades would only be 1.3%.

If the old understanding of the guaranteed Lucky mechanism was flawed, what might the true mechanism be? We assumed, as implied by Niantic’s announcement, that the Lucky guarantee is based on some quantity of Lucky Trades being less than 10, and we set out to determine which Lucky Trades are counted towards this threshold. Based on the observation that the trades from Dawn to May continued to be Lucky even after May had completed a total of ten guaranteed Lucky Trades (the first five sent to Misty and the next five received from Dawn), and because it appeared to agree with common experience,3 we also assumed that the Lucky guarantee is determined entirely by the trading history of the sender of the July-August 2016 Pokémon, with the receiver’s trading history playing no part.

The most natural hypothesis (and the one that was indeed left standing in the end) was that all guaranteed Lucky Trades, rather than all Lucky Trades, count towards the threshold. However, there were a number of other possible criteria for the 10-trade threshold that we felt required investigation before we could be confident in this hypothesis.

Exploring Possibilities

We considered the following alternative Lucky guarantee thresholds:

  1. Ten guaranteed Luckies previously sent (as opposed to previously sent or received).
  2. Ten Lucky Trades counting from the sender’s first guaranteed Lucky sent.
  3. Ten Lucky Trades counting from the sender’s first guaranteed Lucky trade.
  4. Ten Lucky Trades, guaranteed or not, involving July-August 2016 Pokémon.

Possibility (A) was quickly ruled out: May traded two more July-August 2016 Pokémon to Misty. In this case, the first trade became Lucky, but the second did not. While May had previously been involved as sender or receiver in ten guaranteed Lucky Trades, this was only their seventh July-August 2016 Pokémon sent. (For the record, these two trades were made after the first five Guaranteed Lucky trades from Dawn, but before the final five.)

Possibilities (B) and (C) arose from the suggestion that the threshold might still be counting Lucky Trades (including not-necessarily-guaranteed ones), but that the counter might not be triggered until a Trainer was first involved in a guaranteed Lucky Trade. To address these possibilities, we put Ash and Brock to work.

At the beginning of the experiment, Brock traded one July-August 2016 Pokémon with Gary in a guaranteed Lucky Trade. Next, Brock traded newer Pokémon with Gary until they achieved an additional five (non-guaranteed) Lucky Trades, and then Brock sent four guaranteed Lucky Trades to Ash. At this point, Brock had participated in ten Lucky Trades counting from their first guaranteed Lucky Trade as sender, but only five of them guaranteed. Brock then sent five more July-August 2016 Pokémon to Gary, all of which became Lucky. This strongly suggests that these trades had remained guaranteed, providing substantial evidence against possibility (B).

Ash’s trading career began as the receiver of those four guaranteed Lucky Trades from Brock. Ash then traded newer Pokémon (with May and Dawn) until they achieved an additional four (non-guaranteed) Lucky Trades. Ash then sent two July-August 2016 Pokémon to Dawn. At this point Ash had participated in ten Lucky Trades counting from their first guaranteed Lucky Trade as the receiver, but only six them guaranteed. Ash next sent two more July-August 2016 Pokémon to Dawn and three to May; the first four become Lucky, and the fifth did not. This strongly suggests that the first four of this final set of trades had remained guaranteed, providing substantial evidence against possibility (C).

Last, we asked Kiawe to address possibility (D), that Lucky Trades involving July-August 2016 Pokémon might count towards the 10 guaranteed Lucky trades. To do this we took advantage of the fact that Ash still had a supply of July-August 2016 Pokémon, but had exhausted their ten guaranteed Lucky Trades as sender. Kiawe sent one guaranteed Lucky to Ash, and then received thirteen July-August 2016 Pokémon from Ash, of which four became Lucky. At this point Kiawe had been involved in five Lucky Trades involving July-August 2016 Pokémon, only one of which was a guaranteed Lucky Trade. Kiawe then traded ten more July-August 2016 Pokémon to Ash, of which the first nine became Lucky and the tenth did not. Had possibility (D) been in effect, only the first five of those trades (instead of the first nine) would have been guaranteed. That the first nine all became Lucky is substantial evidence against possibility (D).

Open Questions

All of our experiments were strikingly consistent with the hypothesis stated previously: a trade will be guaranteed to be Lucky if the player sending the July-August 2016 Pokémon has previously been involved in fewer than 10 guaranteed Lucky Trades. We are compelled to mention, however, that possibilities (B)-(D) were ruled out on the basis of having observed at least four more consecutive Lucky Trades than one would otherwise expect. Of course the trading trainers might simply have been, well, lucky. The evidence against these possibilities would be bolstered by reproducing these experiments. And there may, of course, be other possibilities that we did not think of testing.

One early hypothesis was that prolific trading might trigger a soft-cap that would lower the success rate of Lucky Trades. While we cannot entirely rule out the existence of soft-caps in some circumstances, Silph Scientists have not observed any evidence of a cap in the data thus far.

Lucky Friends?

While the experiments described here were taking place, Niantic rolled out its Lucky Friends mechanic. We currently do not know whether guaranteed Lucky Trades from the Lucky Friends mechanic will count towards the 10-guaranteed Lucky Trade threshold or not.

Parting Words

The Silph Research Group’s ongoing investigation of Lucky trades has proven to be quite a journey. We appreciate travelers all over the world contributing their anecdotal data, and this episode is a great reminder of why there is always more research to be done! We will continue to gather data on the Lucky rates in order to ascertain an accurate Lucky rate for trades involving July-August 2016 Pokémon as well as soon-to-be 3-year-old Pokémon.

Keep trading and we’ll see you on the road, travelers!

Acknowledgments

Special thanks to Scientist Gluglumaster and Senior Researcher Paleshadow691, as well as our team of volunteer traders and researchers for their extra efforts in unraveling the mysteries of Lucky Pokémon.


1 Many travelers have reported that trades conducted before the September 2018 update to Lucky Trades seem to have counted towards guarantee threshold totals differently. Additionally, these stats from these early trades appear to have been maintained and continue to count towards the guarantee threshold after the update. This adds further complexity to the mix in certain circumstances.

2 Our experiments began before the introduction of the Lucky Friends mechanic, and the participating trainers did not achieve Lucky Friend status with other trainers during the course of the experiment.

3 For clarity’s sake, this “common experience” is the fact that the sender of a July-August 2016 Pokémon during a trade is the one who influences whether or not it can be guaranteed Lucky based on their past interactions.